This paper presents two different recursive prediction error method (RPEM) implementations of multivariate Auto-Regressive Moving-Average (ARMAV) models for identification of a time variant civil engineering structure subject to an earthquake. The two techniques are tested on measurements mode during a pseudo dynamic test performed at the European Laboratory for Structural Assessment (EISA) on a full scale model of a four storey RC-building. Further, the estimated natural frequencies obtained with the two techniques are compared with natural frequencies calculated by using stiffness secant matrices estimated in a previous paper (Rytter et al [14]). The results obtained by the two ARMAV techniques correspond to results obtained using secant matrices.
展开▼